Bengaluru-based voice AI startup Karta AI has raised $3 million in a seed funding round led by PeerCapital and Cedar Hill Capital, with participation from
As AI workloads demand greater memory capacity, chipmakers are exploring alternatives to traditional memory architectures. According to ZDNet, industr...
Mark Zuckerberg wants Meta to monetise its Muse AI agent sooner through transaction fees and subscriptions. His earnings-call answers reveal the costs behind that urgency—and what the model could mean for businesses, including India’s WhatsApp merchants
Global Fishing Watch on October 21 launched its biggest AI overhaul since 2016: new fishing detection and vessel classification models trained on over 1,000 expert-reviewed examples cut false positives. A 2024 Nature study confirms 75% of industrial fishing vessels remain beyond reach of any
Apple plans to launch an updated version of its Siri AI assistant in Korean through a software update as early as 26 October, adding contextual understanding and personalised responses based on information stored on users' devices, according to Chosun, citing company sources. The Korean release follows the launch of an English beta version in September and will coincide with the expansion of language support to Japanese.
One quote from Elon Musk sent chip stocks surging in premarket trading, but whether that momentum signals a real turning point or just a brief relief rally is what the rest of October will decide.
Full 3D attention is costly in video Diffusion Transformers. Block-sparse attention reduces this cost by computing only selected attention blocks. However, existing methods either overlook the potential to optimize sparsity allocation or rely on heuristic rules or local metrics…
Jet tagging is one of the most important applications of machine learning at the Large Hadron Collider. Recent models build in more physics knowledge or rely on large-scale pretraining, but the experiments place strict limits on their inference cost. Knowledge distillation can…
World models are increasingly used for planning, where a planner ranks candidate actions by returns imagined by the model. However, these models are judged mainly by how well they render frames, even though a visually faithful rollout can still assign the wrong return. To…
A question about a whole-slide image rarely names every morphological finding needed to answer it. Question-guided patch selection can therefore overlook supporting morphology, while slide-level compression can dilute focal evidence. Inspired by how pathologists combine broad…